Coqui TTS command line help

readme.md

Retrieved with:

MSYS_NO_PATHCONV=1 docker run --rm --entrypoint /bin/bash ghcr.io/coqui-ai/tts -o errexit -o xtrace -c 'tts --help; python3 /root/TTS/server/server.py --help; python3 /root/TTS/bin/train_tts.py --help'

serverHelp.txt

usage: server.py [-h] [--list_models [LIST_MODELS]] [--model_name MODEL_NAME] [--vocoder_name VOCODER_NAME] [--config_path CONFIG_PATH] [--model_path MODEL_PATH] [--vocoder_path VOCODER_PATH]
                 [--vocoder_config_path VOCODER_CONFIG_PATH] [--speakers_file_path SPEAKERS_FILE_PATH] [--port PORT] [--use_cuda USE_CUDA] [--debug DEBUG] [--show_details SHOW_DETAILS]

options:
  -h, --help            show this help message and exit
  --list_models [LIST_MODELS]
                        list available pre-trained tts and vocoder models.
  --model_name MODEL_NAME
                        Name of one of the pre-trained tts models in format <language>/<dataset>/<model_name>
  --vocoder_name VOCODER_NAME
                        name of one of the released vocoder models.
  --config_path CONFIG_PATH
                        Path to model config file.
  --model_path MODEL_PATH
                        Path to model file.
  --vocoder_path VOCODER_PATH
                        Path to vocoder model file. If it is not defined, model uses GL as vocoder. Please make sure that you installed vocoder library before (WaveRNN).
  --vocoder_config_path VOCODER_CONFIG_PATH
                        Path to vocoder model config file.
  --speakers_file_path SPEAKERS_FILE_PATH
                        JSON file for multi-speaker model.
  --port PORT           port to listen on.
  --use_cuda USE_CUDA   true to use CUDA.
  --debug DEBUG         true to enable Flask debug mode.
  --show_details SHOW_DETAILS
                        Generate model detail page.

trainHelp.txt

usage: train_tts.py [-h] [--continue_path CONTINUE_PATH] [--restore_path RESTORE_PATH] [--best_path BEST_PATH] [--use_ddp true/false] [--use_accelerate true/false] [--grad_accum_steps GRAD_ACCUM_STEPS]
                    [--overfit_batch true/false] [--skip_train_epoch true/false] [--start_with_eval true/false] [--small_run SMALL_RUN] [--gpu GPU] [--rank RANK] [--group_id GROUP_ID] [--config_path CONFIG_PATH]

options:
  -h, --help            show this help message and exit
  --continue_path CONTINUE_PATH
                        Coqpit Field: Path to a training folder to continue training. Restore the model from the last checkpoint and continue training under the same folder.
  --restore_path RESTORE_PATH
                        Coqpit Field: Path to a model checkpoit. Restore the model with the given checkpoint and start a new training.
  --best_path BEST_PATH
                        Coqpit Field: Best model file to be used for extracting the best loss. If not specified, the latest best model in continue path is used
  --use_ddp true/false  Coqpit Field: Use DDP in distributed training. It is to set in `distribute.py`. Do not set manually.
  --use_accelerate true/false
                        Coqpit Field: Use HF Accelerate as the back end for training.
  --grad_accum_steps GRAD_ACCUM_STEPS
                        Coqpit Field: Number of gradient accumulation steps. It is used to accumulate gradients over multiple batches.
  --overfit_batch true/false
                        Coqpit Field: Overfit a single batch for debugging.
  --skip_train_epoch true/false
                        Coqpit Field: Skip training and only run evaluation and test.
  --start_with_eval true/false
                        Coqpit Field: Start with evaluation and test.
  --small_run SMALL_RUN
                        Coqpit Field: Only use a subset of the samples for debugging. Set the number of samples to use. Defaults to None.
  --gpu GPU             Coqpit Field: GPU ID to use if ```CUDA_VISIBLE_DEVICES``` is not set. Defaults to None.
  --rank RANK           Coqpit Field: Process rank in a distributed training. Don't set manually.
  --group_id GROUP_ID   Coqpit Field: Process group id in a distributed training. Don't set manually.
  --config_path CONFIG_PATH
                        Coqpit Field: Path to the config file.

ttsHelp.txt

2023-08-05 15:35:03 usage: tts [-h] [--list_models [LIST_MODELS]]
2023-08-05 15:35:03            [--model_info_by_idx MODEL_INFO_BY_IDX]
2023-08-05 15:35:03            [--model_info_by_name MODEL_INFO_BY_NAME] [--text TEXT]
2023-08-05 15:35:03            [--model_name MODEL_NAME] [--vocoder_name VOCODER_NAME]
2023-08-05 15:35:03            [--config_path CONFIG_PATH] [--model_path MODEL_PATH]
2023-08-05 15:35:03            [--out_path OUT_PATH] [--use_cuda USE_CUDA]
2023-08-05 15:35:03            [--vocoder_path VOCODER_PATH]
2023-08-05 15:35:03            [--vocoder_config_path VOCODER_CONFIG_PATH]
2023-08-05 15:35:03            [--encoder_path ENCODER_PATH]
2023-08-05 15:35:03            [--encoder_config_path ENCODER_CONFIG_PATH] [--emotion EMOTION]
2023-08-05 15:35:03            [--speakers_file_path SPEAKERS_FILE_PATH]
2023-08-05 15:35:03            [--language_ids_file_path LANGUAGE_IDS_FILE_PATH]
2023-08-05 15:35:03            [--speaker_idx SPEAKER_IDX] [--language_idx LANGUAGE_IDX]
2023-08-05 15:35:03            [--speaker_wav SPEAKER_WAV [SPEAKER_WAV ...]]
2023-08-05 15:35:03            [--gst_style GST_STYLE]
2023-08-05 15:35:03            [--capacitron_style_wav CAPACITRON_STYLE_WAV]
2023-08-05 15:35:03            [--capacitron_style_text CAPACITRON_STYLE_TEXT]
2023-08-05 15:35:03            [--list_speaker_idxs [LIST_SPEAKER_IDXS]]
2023-08-05 15:35:03            [--list_language_idxs [LIST_LANGUAGE_IDXS]]
2023-08-05 15:35:03            [--save_spectogram SAVE_SPECTOGRAM] [--reference_wav REFERENCE_WAV]
2023-08-05 15:35:03            [--reference_speaker_idx REFERENCE_SPEAKER_IDX]
2023-08-05 15:35:03            [--progress_bar PROGRESS_BAR] [--source_wav SOURCE_WAV]
2023-08-05 15:35:03            [--target_wav TARGET_WAV] [--voice_dir VOICE_DIR]
2023-08-05 15:35:03 
2023-08-05 15:35:03 Synthesize speech on command line.
2023-08-05 15:35:03 
2023-08-05 15:35:03 You can either use your trained model or choose a model from the provided list.
2023-08-05 15:35:03 
2023-08-05 15:35:03 If you don't specify any models, then it uses LJSpeech based English model.
2023-08-05 15:35:03 
2023-08-05 15:35:03 ## Example Runs
2023-08-05 15:35:03 
2023-08-05 15:35:03 ### Single Speaker Models
2023-08-05 15:35:03 
2023-08-05 15:35:03 - List provided models:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --list_models
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Query info for model info by idx:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --model_info_by_idx "<model_type>/<model_query_idx>"
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Query info for model info by full name:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --model_info_by_name "<model_type>/<language>/<dataset>/<model_name>"
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run TTS with default models:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --text "Text for TTS"
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run a TTS model with its default vocoder model:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --text "Text for TTS" --model_name "<model_type>/<language>/<dataset>/<model_name>
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run with specific TTS and vocoder models from the list:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --text "Text for TTS" --model_name "<model_type>/<language>/<dataset>/<model_name>" --vocoder_name "<model_type>/<language>/<dataset>/<model_name>" --output_path
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run your own TTS model (Using Griffin-Lim Vocoder):
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --text "Text for TTS" --model_path path/to/model.pth --config_path path/to/config.json --out_path output/path/speech.wav
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run your own TTS and Vocoder models:
2023-08-05 15:35:03     $ tts --text "Text for TTS" --model_path path/to/config.json --config_path path/to/model.pth --out_path output/path/speech.wav
2023-08-05 15:35:03         --vocoder_path path/to/vocoder.pth --vocoder_config_path path/to/vocoder_config.json
2023-08-05 15:35:03 
2023-08-05 15:35:03 ### Multi-speaker Models
2023-08-05 15:35:03 
2023-08-05 15:35:03 - List the available speakers and choose as <speaker_id> among them:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --model_name "<language>/<dataset>/<model_name>"  --list_speaker_idxs
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run the multi-speaker TTS model with the target speaker ID:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --text "Text for TTS." --out_path output/path/speech.wav --model_name "<language>/<dataset>/<model_name>"  --speaker_idx <speaker_id>
2023-08-05 15:35:03 
2023-08-05 15:35:03 - Run your own multi-speaker TTS model:
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --text "Text for TTS" --out_path output/path/speech.wav --model_path path/to/config.json --config_path path/to/model.pth --speakers_file_path path/to/speaker.json --speaker_idx <speaker_id>
2023-08-05 15:35:03 
2023-08-05 15:35:03 ### Voice Conversion Models
2023-08-05 15:35:03 
2023-08-05 15:35:03     $ tts --out_path output/path/speech.wav --model_name "<language>/<dataset>/<model_name>" --source_wav <path/to/speaker/wav> --target_wav <path/to/reference/wav>
2023-08-05 15:35:03     
2023-08-05 15:35:03 
2023-08-05 15:35:03 options:
2023-08-05 15:35:03   -h, --help            show this help message and exit
2023-08-05 15:35:03   --list_models [LIST_MODELS]
2023-08-05 15:35:03                         list available pre-trained TTS and vocoder models.
2023-08-05 15:35:03   --model_info_by_idx MODEL_INFO_BY_IDX
2023-08-05 15:35:03                         model info using query format: <model_type>/<model_query_idx>
2023-08-05 15:35:03   --model_info_by_name MODEL_INFO_BY_NAME
2023-08-05 15:35:03                         model info using query format: <model_type>/<language>/<dataset>/<model_name>
2023-08-05 15:35:03   --text TEXT           Text to generate speech.
2023-08-05 15:35:03   --model_name MODEL_NAME
2023-08-05 15:35:03                         Name of one of the pre-trained TTS models in format <language>/<dataset>/<model_name>
2023-08-05 15:35:03   --vocoder_name VOCODER_NAME
2023-08-05 15:35:03                         Name of one of the pre-trained  vocoder models in format <language>/<dataset>/<model_name>
2023-08-05 15:35:03   --config_path CONFIG_PATH
2023-08-05 15:35:03                         Path to model config file.
2023-08-05 15:35:03   --model_path MODEL_PATH
2023-08-05 15:35:03                         Path to model file.
2023-08-05 15:35:03   --out_path OUT_PATH   Output wav file path.
2023-08-05 15:35:03   --use_cuda USE_CUDA   Run model on CUDA.
2023-08-05 15:35:03   --vocoder_path VOCODER_PATH
2023-08-05 15:35:03                         Path to vocoder model file. If it is not defined, model uses GL as vocoder. Please make sure that you installed vocoder library before (WaveRNN).
2023-08-05 15:35:03   --vocoder_config_path VOCODER_CONFIG_PATH
2023-08-05 15:35:03                         Path to vocoder model config file.
2023-08-05 15:35:03   --encoder_path ENCODER_PATH
2023-08-05 15:35:03                         Path to speaker encoder model file.
2023-08-05 15:35:03   --encoder_config_path ENCODER_CONFIG_PATH
2023-08-05 15:35:03                         Path to speaker encoder config file.
2023-08-05 15:35:03   --emotion EMOTION     Emotion to condition the model with. Only available for 🐸Coqui Studio models.
2023-08-05 15:35:03   --speakers_file_path SPEAKERS_FILE_PATH
2023-08-05 15:35:03                         JSON file for multi-speaker model.
2023-08-05 15:35:03   --language_ids_file_path LANGUAGE_IDS_FILE_PATH
2023-08-05 15:35:03                         JSON file for multi-lingual model.
2023-08-05 15:35:03   --speaker_idx SPEAKER_IDX
2023-08-05 15:35:03                         Target speaker ID for a multi-speaker TTS model.
2023-08-05 15:35:03   --language_idx LANGUAGE_IDX
2023-08-05 15:35:03                         Target language ID for a multi-lingual TTS model.
2023-08-05 15:35:03   --speaker_wav SPEAKER_WAV [SPEAKER_WAV ...]
2023-08-05 15:35:03                         wav file(s) to condition a multi-speaker TTS model with a Speaker Encoder. You can give multiple file paths. The d_vectors is computed as their average.
2023-08-05 15:35:03   --gst_style GST_STYLE
2023-08-05 15:35:03                         Wav path file for GST style reference.
2023-08-05 15:35:03   --capacitron_style_wav CAPACITRON_STYLE_WAV
2023-08-05 15:35:03                         Wav path file for Capacitron prosody reference.
2023-08-05 15:35:03   --capacitron_style_text CAPACITRON_STYLE_TEXT
2023-08-05 15:35:03                         Transcription of the reference.
2023-08-05 15:35:03   --list_speaker_idxs [LIST_SPEAKER_IDXS]
2023-08-05 15:35:03                         List available speaker ids for the defined multi-speaker model.
2023-08-05 15:35:03   --list_language_idxs [LIST_LANGUAGE_IDXS]
2023-08-05 15:35:03                         List available language ids for the defined multi-lingual model.
2023-08-05 15:35:03   --save_spectogram SAVE_SPECTOGRAM
2023-08-05 15:35:03                         If true save raw spectogram for further (vocoder) processing in out_path.
2023-08-05 15:35:03   --reference_wav REFERENCE_WAV
2023-08-05 15:35:03                         Reference wav file to convert in the voice of the speaker_idx or speaker_wav
2023-08-05 15:35:03   --reference_speaker_idx REFERENCE_SPEAKER_IDX
2023-08-05 15:35:03                         speaker ID of the reference_wav speaker (If not provided the embedding will be computed using the Speaker Encoder).
2023-08-05 15:35:03   --progress_bar PROGRESS_BAR
2023-08-05 15:35:03                         If true shows a progress bar for the model download. Defaults to True
2023-08-05 15:35:03   --source_wav SOURCE_WAV
2023-08-05 15:35:03                         Original audio file to convert in the voice of the target_wav
2023-08-05 15:35:03   --target_wav TARGET_WAV
2023-08-05 15:35:03                         Target audio file to convert in the voice of the source_wav
2023-08-05 15:35:03   --voice_dir VOICE_DIR
2023-08-05 15:35:03                         Voice dir for tortoise model